5 MATRICES 371
pA are linearly independent over K'. Hence the p^ba are linearly independent
over K', and our assertion is proved. In particular:

(A.4.21) If E is finite dimensional over K, and K is finite dimensional over K',
then E is finite dimensional over K' and we have

(A.4.21.1) dimK, E = dimK E • dimK, K.
5. MATRICES
(A.5.1) Let E, F be vector spaces over a field K, and suppose that E is of
finite dimension n. Let (at-)1^I-^n be a basis of E, so that E is the direct sum
of the Kat. By (A.3.1) there is a one-one correspondence between the linear
mappings u of E into F and the families (b^^t^ of n vectors of F; this
correspondence is defined by bt = u(at) (1 ^ i < n). Thus the vector space
Hom(E, F) is isomorphic to Fn.

(A.5.2) Suppose furthermore that F is of finite dimension m, and let
(bj)i^jt$m be a basis of F. Then there is a one-one correspondence between
the vectors y e F and the families (^j)i^j^m °f m elements of K, defined by

y= £*?/&/. Hence there is a one-one correspondence between the linear
j«i
mappings u : E -* F and the " double " families (o^) (with 1 <y < w, 1 < / < «)

of elements of K, defined by the relations
Such families are called matrices with m rows and n columns (or m x n
matrices)
over K. They form a vector space over K, isomorphic to Kmn. The
subfamily (o,,)^^ is theyth row, and the subfamily (a/Di*^™ is the zth
column, of the matrix (ajj)1<</iSWfl<j<B. The matrix M(u) == (o^) defined by
(A.5.2.1) is called the matrix ofu with respect to the bases (at) and (&,-). If w, u'
are two linear mappings of E into F, and if 1 is any scalar, then

M(u + iO = M(u) + M(u'\
M(lu)
= AAf («),

the matrices being taken in each case with respect to the same bases (at) of E
and (bj) of F.